You need to know more about the data. Otherwise you will need to use Chebyshev's inequality - Wikipedia, the free encyclopedia
Suppose we have a set of blood pressures with a mean of 120 Systolic, and a sample standard deviation of 20 points. If we cannot assume a normal distribution of Systolic blood pressures, then between what two values can we be assured 99.7% of all Systolic blood pressures will lie?
????
You need to know more about the data. Otherwise you will need to use Chebyshev's inequality - Wikipedia, the free encyclopedia